目的:評估骨盆固定器對於放射治療擺位誤差的影響。10例，攝護腺癌4例，及直腸癌2例。在執行放射治療計畫之前即以隨機抽樣的方式區分為傳統未加骨盆固定器與使用骨盆固定器兩組患者。開始治療後每周拍攝兩張照野片，照野片拍攝的方向是前後或兩側面方向。本研究共取得34張模擬攝影片與149張照野片。放射治療方式在兩組之間沒有任何差異，除直腸癌是以三照野治療外其餘均是四照野方式治療。記錄放射治療擺位誤差的方法為比較模擬攝影X光片與照野片的左右、頭腳、前後、以及旋轉等軸的誤差。分析擺位誤差的方法是以t-test(兩尾)及chi-square方法分別評估誤差平均值及誤差超過5mm比例等數值有無統計學上的差異。均值方式評估，兩者左右(p=0.445)、頭腳(p=0.092)、與旋轉軸(p=0.337)均未有明顯差別。唯有機頭270度時的前後軸達到統計差異(p=0.016);使用骨盆固定器患者的平均誤差(0.21mm, SD=1.11mm)明顯小於未使用骨盆固定器患者的平均誤差(1.54mm, SD=2.58mm)。以誤差超過5mm比例的方法作評估，使用骨盆固定器可減少前後軸的誤差(p=0.046)，但反而使頭腳軸的誤差加大(p=0.018);左右軸則未有顯著差別(p=0.359)例而言它反而使頭腳軸方向的誤差加大;對於左右軸方向的誤差並未有明顯的影響。所以當使用骨盆固定器時應注意頭腳軸方向的固定方法以減少此方向軸的誤差。

　　　　　Purpose: To evaluate the extent of influence of setup deviation by pelvicimmobilization device.groups that were with or without pelvic immobilization device. We included three kindsof malignant diseases that originated in pelvic cavity; these were 10 cases ofcervical cancer, 4 cases of prostate cancer, and 2 cases of rectal cancer. Exceptfrom the two patients of rectal cancer, all of them were treated with four-field boxtechnique and the principle of radiation technique was all the same in both groups.The portal films were taken from anterior-posterior and bilateral directions withthe frequency of twice per week. There were 34 simulation films and 149 portal filmsto be evaluated. The setup error was recorded by comparison the iso-center shiftbetween simulation and portal films of four individual axes, that were latero-lateral,caudo-cranial, anterior-posterior axis and degree of rotation. The statistic methodsfor evaluation the results were two-tail t-test and chi-square methods for mean valueof setup deviation and ratio of iso-center shift exceeded 5-mm. respectively.and immobilization groups. When evaluated by mean setup error, there were nosignificant differencer between the two groups of the latero-lateral (p=0.445) andcaudo-cranial (p=0.092) axes, the degree of rotation was also no difference at all(p=0.337). In anterior-posterior axis evaluated by gantry 270 degree, the mean setuperror of the immobilization group (0.21-mm, SD=1.11mm) was significant smaller thannon-immobilization group 1.54-mm, SD=2.58mm) (p=0.016). Evaluation by the ratio ofiso-center shift exceeded 5-mm, the immobilized group showed significant reductionin the error of anterior-posterior axis (p=0.046), but this group had opposed behaviorof the caudo-cranial axis (p=0.018). The latero-lateral (p=0.359) and rotational(p=0.803) axes did not show any significant difference between the two groups.of the anterior-posterior axis, but the caudo-cranial axis has opposed effect whenwe evaluated by the ratio of setup error exceeded 5-mm. The iso-center shift inlatero-lateral axis has no significant difference despite of the used of thisimmobilization device. So, we must pay more effort to avoid the inter-treatmentvariation of the caudo-cranial axis when we use this kind of pelvic immobilizationdevice.